Across the world's leading economies, a new kind of economic duality is taking hold. Headline growth figures look reassuring — but beneath the surface, the engine driving them is increasingly narrow, concentrated in the data centres, GPU clusters, and fibre corridors of artificial intelligence infrastructure.
The latest evidence comes from the United States, where CoreWeave, a GPU-specialised cloud provider, has closed a $1.17 billion deal that cements its role as a barometer for enterprise AI compute demand. Combined with SoftBank's reported pursuit of a Marvell Technology acquisition, the transaction illustrates how capital is flowing into AI at a pace and scale that is visibly moving top-line economic indicators — even as the broader population experiences something quite different.
A GDP Story With Two Authors
US second-quarter GDP growth came in at 3.8%, with a similar reading expected for the third quarter. But former Federal Reserve Governor Lael Brainard has been candid about what is writing that number: "The economy at the top level is strong, but again, it's being driven by this really important set of investments in AI. The rest of the economy under the hood is really stuck."
That diagnosis will resonate far beyond Washington. In the United Kingdom, Germany, France, and Japan, policymakers are grappling with analogous tensions: aggregate growth figures buoyed by technology investment, while manufacturing, retail, and lower-income households face an entirely different reality. The International Monetary Fund has repeatedly flagged that productivity gains from AI are likely to accrue unevenly, both within countries and between them, with advanced economies capturing the bulk of near-term benefits.
The Global Central Bank Dilemma
The Federal Reserve's predicament is, in structural terms, a global one. With inflation still running at roughly 3% in the United States — partly fuelled by tariff effects — the FOMC has signalled that the bar for further rate cuts is high. Fed Governor Philip Jefferson has described current rates as sitting within a "neutral range," a formulation that masks a deeper problem: monetary policy calibrated for the average of two diverging economies satisfies neither.
The European Central Bank faces an analogous bind. The eurozone's AI investment wave is smaller in absolute terms but growing rapidly, with Germany's industrial giants and France's emerging AI ecosystem attracting substantial sovereign and private capital. Yet the ECB must balance those pockets of dynamism against a broader continental economy where consumer confidence remains fragile and energy costs elevated.
In Asia, the calculus differs again. China is pursuing AI infrastructure expansion through a combination of state direction and private investment, with domestic champions such as Huawei and Baidu building compute capacity partly insulated from Western capital markets. Japan and South Korea, meanwhile, are investing heavily in semiconductor and data centre infrastructure as strategic national priorities, with government subsidies blurring the line between market-driven and policy-driven growth.
AI Infrastructure as a Distinct Global Asset Class
What the CoreWeave deal makes plain is that AI infrastructure has matured into a recognised asset class — one with characteristics that distinguish it sharply from conventional enterprise technology spending. GPU cluster procurement, long-duration data centre leases, and energy supply agreements create durable demand signals that ripple outward to chip manufacturers, grid operators, construction firms, and commercial real estate across multiple continents.
The geographic footprint of that ripple is significant. NVIDIA's supply chain spans Taiwan, South Korea, the Netherlands, and Malaysia. Data centre construction is driving land acquisition and power infrastructure investment from Ireland to Singapore. The financial flows involved in a single deal of CoreWeave's scale touch more jurisdictions than almost any other category of private investment.
Who Is Left Behind — And Where
The structural concern that former Atlanta Fed President Dennis Lockhart articulated — that central banks may be forced to remain data-dependent even as that data reflects two simultaneous economies — applies with particular force internationally. In emerging markets, the AI investment boom is largely happening to them rather than with them. Countries in Sub-Saharan Africa, South Asia, and Latin America are seeing capital concentrated elsewhere, even as they bear some of the inflationary and exchange-rate consequences of tighter monetary conditions in the developed world.
The IMF has urged advanced economies to consider the spillover effects of their AI-driven investment surges, warning that widening productivity gaps could entrench structural disadvantages for lower-income countries that lack the infrastructure, talent pipelines, or capital markets to participate in the current cycle.
The Policy Question No One Has Answered
What distinguishes this moment from previous technology investment cycles — the dotcom era, the mobile buildout, the cloud transition — is the speed and concentration of capital deployment, and the degree to which it is now large enough to distort macroeconomic readings in the world's largest economy. When AI infrastructure spending can move a country's GDP figure by a measurable margin, it ceases to be a sectoral story and becomes a systemic one.
Governments and multilateral institutions are only beginning to develop frameworks adequate to that reality. The question is not whether AI infrastructure investment will continue — the demand signals, from hyperscaler capex commitments to sovereign AI strategies, suggest it will accelerate. The question is whether the institutions designed to manage economic stability were built for a world in which a single technology investment cycle can simultaneously inflate a headline number and hollow out the economy it is supposed to represent.
Sources:
1 Yahoo Finance, "Fed fallout, missing jobs numbers, and a busy earnings calendar: What to watch this week" (November 02, 2025)
2 Yahoo Finance, "Fed is in 'unusual juncture' on rates, lack of data: Lael Brainard" (November 13, 2025)
3 Yahoo Finance, "'I think time is running out': How to survive a market bubble — if there is one" (December 24, 2025)
4 Nasdaq, "Software Stocks Retreat and Drag the Broader Market Lower" (February 03, 2026)
5 Yahoo Finance, "Stock market today: Dow closes above 50,000 for the first time as stocks soar to cap volatile week" (February 06, 2026)